Test of Block Zero Restrictions in Factor Loadings

다국어 초록 (Multilingual Abstract)

As a companion paper to Han and Kim (2019) on testing for block zero restrictions in common factors, this paper proposes a test of block zero restrictions in factor loadings, which have empirical relevance in various topics in economics. The test statistic is constructed using the principal component estimate of factor loadings and has a chi-square distribution asymptotically under the null hypothesis of block zero restrictions. The test is applied to a panel of 132 monthly U.S. macro economic time series, which are classified into three groups. The test results show a multi-level structure in factor loadings.

참고문헌 (Reference)

1 Han, C., "Testing for the Null of Block Zero Restrictions in Common Factor Models, Manuscript"

2 Stock, J.H, "Macroeconomic Forecasting Using Diffusion Indexes" 20 : 147-162, 2002

3 Ludvigson, S.C., "Macro Factors in Bond Risk Premia" 22 : 5027-5067, 2009

4 Bai, J., "Inferential Theory for Factor Models of Large Dimensions" 71 : 135-172, 2003

5 Andrews, D. W. K., "Heteroskedasticity and Autocorrelation Consistent Covariance Matrix Estimation" 59 : 817-858, 1991

6 Hallin, M., "Dynamic Factors in the Presence of Blocks" 163 : 29-21, 2011

7 Dias, F., "Determining the Number of Global and Country-Specific Factors in the Euro Area" 17 : 573-517, 2013

8 Bai, J., "Determining the Number of Factors in Approximate Factor Models" 70 : 191-121, 2002

9 Choi, I., "A Multi-level Factor Model : Identification, Asymptotic Theory and Applications" 33 : 355-377, 2018